If your need is to produce text in a language other than English in your story, then the best ap is an actual live interpreter/translator. Translator aps do actually function when you feed them hyper-precise input text. I can pass paragraphs of Westlaw's legal database through Google Translate, and the results are surprisingly good, but the only reason for that outcome is the exhaustively painstaking, pedantry-as-art-form, perfectionist manner over which case law is pored before it is written. Not one single tiny vagary can be present, not a single instance of a thing that could remotely have more than one meaning. Unless you have that kind of skill and are willing to do that first, there are tons of interpreter/translator websites. We take all kinds of jobs. This is what I do for a living if I've not made that clear.
Writer, "They're all equally bad." Google translator, "Necesitamos más Pledge de Limón" See? Google translator works great.
Adding this, I used to live translate between three languages. If translation software worked as intended, there wouldn't have been a job for me. The most frequent advertisement you'll see from translation agencies will one way or another cover "software translation briefly looked over by one of our fellows".
Brilliantly played! To stay on topic, I think there's a Carter-era translator still looking for work - https://www.mentalfloss.com/article/19249/president-jimmy-carters-carnal-mistake If I understand this phrasebook correctly, his hovercraft is full of eels.
And they'll only do that with European languages. It's not just the difference in thought systems, there's a serious gap in the programming. Google translate, whatever FB uses, and one other that I know of but can't think of still translate the Japanese for things like "21st" as "21th." It's one of the cues we look for when spotting machine translation. Japanese and Korean also don't use subjects, so things like "I like cheese" and "He likes cheese," "we like cheese" all come out the same when going from proper, grammatical Japanese (チーズすきです)to English.
Same with the pro-drop nature of Spanish that gives it seizures. The conjugation paradigm for Spanish verbs is as robust and conservative as they come. The verb carries all the information you need to know subject vs object. Yet the simple lack of an explicit subject pronoun throws GT's "who did what to whom" into chaos. It does make for hilarity when we play those "pass your paragraph through GT a couple of times and then post the results" games.
Unfortunately, Google is probably the best. Most just use lookup tables, but Google's actually has a deep learning component. They feed it official translations from organizations like the UN. This has fixed a lot of problems like phrases and idioms not properly translating, but it still doesn't really understand what is being said, it's all a categorization network.
Thanks. I checked out. It looks doesn’t support most/ all languages. Any way to add a language that isn’t on the list.
No, DeepL supports only English (British and U.S.), German, French, Spanish, Portuguese, Italian, Dutch, Polish, Russian, Japanese, and Chinese (Simplified). The only way to add a language would be to petition the developers and ask if/when they might consider adding ___ to the mix.
Thanks. It seems most of the languages it supports are european languages. Will contact them to see if there is a way of adding a new language.